Multi-Layer Codec Quantization Parameter Reuse

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing scalable video coding techniques do not effectively reuse parameters related to quantization of transform coefficient data across layers, which hinders coding efficiency in multi-layer codecs, including scalable and multi-view codecs.

Innovation Solution

The solution involves reusing scaling lists and quantization parameters across layers in a multi-layer codec, allowing for optimized quantization and inverse quantization processes by sharing common parameters between layers, such as through the use of quantization matrixes and offset calculations.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If quantization parameters are encoded separately in each layer, then each layer can be independently optimized, but coding efficiency deteriorates due to redundant parameter encoding

Engineering Contradiction:
Improvecoding efficiencyVSAvoidparameter encoding redundancy
Core Design Contradiction:
ProductivityVSQuantity of substance

Solution Approach 1:

The patent merges the encoding of quantization parameters across multiple layers by introducing inter-layer prediction mechanisms. The base layer quantization parameters are encoded normally, while enhancement layer parameters are predicted from base layer parameters using prediction modes (copy mode, offset mode, or combination mode), thereby combining the parameter representation across layers and eliminating redundancy.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent creates a universal quantization parameter representation that serves multiple layers simultaneously. By establishing a base layer parameter set that can be reused and adapted for enhancement layers through prediction, the system achieves multi-functionality where a single parameter encoding serves the purpose of multiple layer quantization, improving overall coding efficiency.

Inventive Principle:
Principle #6Universality (Multi-functionality)

2Productivity

If quantization parameters are reused across layers, then coding efficiency improves by reducing redundancy, but adaptability deteriorates as layer-specific optimizations are lost

Engineering Contradiction:
Improvecoding efficiencyVSAvoidlayer-specific optimization capability
Core Design Contradiction:
ProductivityVSAdaptability or versatility

Solution Approach 1:

The patent introduces dynamic adaptability through multiple prediction modes that can be selected based on layer characteristics. The system can switch between copy mode (full reuse), offset mode (partial adaptation), and combination mode (hybrid approach), allowing the quantization parameter prediction mechanism to adapt dynamically to different layer requirements while maintaining the overall efficiency benefit of parameter reuse.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The patent applies local quality by allowing different prediction modes to be used for different quantization parameter groups or different enhancement layers. This enables fine-grained control where parameters that benefit from reuse can be copied, while parameters requiring layer-specific optimization can use offset or combination modes, thus maintaining both efficiency and adaptability at appropriate levels.

Inventive Principle:
Principle #3Local quality

3Manufacturing precision

If multiple quantization parameter sets are maintained for different layers, then layer-specific quality control is improved, but device complexity increases

Engineering Contradiction:
Improvelayer-specific quality controlVSAvoidparameter management complexity
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The patent segments the quantization parameter management into a hierarchical structure with base layer parameters and enhancement layer parameters. The base layer parameters serve as a common reference, while enhancement layer parameters are derived through prediction modes. This segmentation reduces complexity by avoiding complete independent management of all parameters while maintaining layer-specific control through the prediction mechanism.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces prediction modes as an intermediary mechanism between base layer parameters and enhancement layer parameters. Instead of directly managing all parameters independently, the system uses prediction modes (copy, offset, combination) as intermediaries that translate base layer parameters into enhancement layer parameters, thereby simplifying the overall parameter management complexity while preserving layer-specific quality control.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP3273691B1Image processing device and image processing method
Publication Date: 2021.09.22 SONY GROUP CORP
  • EP3273691B1 patent drawingFigure 1
  • EP3273691B1 patent drawingFigure 2
  • EP3273691B1 patent drawingFigure 3

AI summary

Provided is an image processing device including a control section configured to set, based on a first quantization parameter offset set for a chroma component of a first layer, a second quantization parameter offset for a chroma component of a second layer decoded with reference to the first layer, and an inverse quantization section configured to inversely quantize transform coefficient data of the chroma component of the second layer using a quantization parameter computed using the second quantization parameter offset set by the control section.